Something interesting is happening in the AI world that's worth paying attention to. We're now starting to see a shift from incremental improvements to something more fundamental in how AI is being built and used.
I’ve been working with small models these past couple of weeks putting them through their paces. Models such as Qwen3 (with 4 billion parameters) or Gemma3 (with a mere 1 billion parameters), all running locally on small, inexpensive hardware without a need for an internet connection, and with complete privacy baked in.
What’s really caught my attention when working with these models is just how capable they are when given proper instructions.
Why does this matter? Because suddenly, capable, usable and helpful AI doesn't need massive server farms. A small business could run game changing AI locally on ordinary hardware. That's not just more efficient – it's democratising technology that was previously only available to tech giants.
Google's Gemini Diffusion is processing text at 1,479 tokens per second – nearly 10 times faster than GPT-4o. This isn't just about quicker responses; it opens doors to real-time applications we couldn't consider before. Think about it: customer service that responds instantly or content creation that feels genuinely conversational. When speed stops being a limitation, new possibilities emerge.
If you're working in digital transformation, here's what I'm watching:
The opportunity: AI capabilities that once required serious infrastructure investment are becoming accessible to businesses of any size. This democratisation is happening now, not someday.
The challenge: The pace is relentless. What seemed advanced six months ago might already be behind the curve. Staying current isn't just about competitive advantage – it's about remaining relevant.
We're moving towards AI as infrastructure – like electricity or broadband. It's shifting from being a nice-to-have tool to essential business infrastructure.
The organisations that will thrive are those experimenting now, adapting quickly, and thinking carefully about how AI fits into their core operations.
The question isn't whether AI will change your industry – it's whether you'll be leading that change or catching up later.